ERIC Number: EJ994001
Record Type: Journal
Publication Date: 2012
Pages: 5
Abstractor: ERIC
ISBN: N/A
ISSN: ISSN-1536-6367
EISSN: N/A
Old Issues in a New Jacket: Power and Validation in the Context of Mixture Modeling
Lubke, Gitta
Measurement: Interdisciplinary Research and Perspectives, v10 n4 p212-216 2012
Von Davier et al. (this issue) describe two analyses that aim at determining whether the constructs measured with a number of observed items are categorical or continuous in nature. The issue of types versus traits has a long history and is relevant in many areas of behavioral research, including personality research, as emphasized by von Davier et al. Types can be investigated using latent class models, whereas traits are modeled with latent trait or factor models. As described in von Davier et al., general modeling frameworks have been proposed that permit a comparison of these models based on fit indices. Ideally then, the decision concerning types versus traits can be facilitated by the results of such a comparison. The analyses presented by von Davier et al. nicely illustrate the difficulties encountered when analyzing empirical data. Two basic issues in mixture modeling, namely the power to discriminate between alternative models and the need to validate results, deserve more emphasis. (Contains 3 tables.)
Descriptors: Models, Classification, Multivariate Analysis, Statistical Analysis, Validity, Goodness of Fit, Bayesian Statistics, Sample Size, Comparative Analysis
Psychology Press. Available from: Taylor & Francis, Ltd. 325 Chestnut Street Suite 800, Philadelphia, PA 19106. Tel: 800-354-1420; Fax: 215-625-2940; Web site: http://www.tandf.co.uk/journals
Publication Type: Journal Articles; Opinion Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A
Grant or Contract Numbers: N/A